Jorick van Weelie
Marketing Lead at DataNorth | AI Enthusiast & Tech Storyteller
Articles written by Jorick van Weelie
OpenAI updates GPT-5.6 Sol and GPT-5.6 Luna
GPT-5.6 Sol in ChatGPT now answers more directly and, by OpenAI's own measurement, makes about 68% fewer factual errors than GPT-5.5 Instant, while free users move to GPT-5.6 Luna with
Google DeepMind open sources WeatherNext 2 and WeatherNext Cyclones
Google DeepMind has open sourced WeatherNext 2, WeatherNext 2-mini and WeatherNext Cyclones under commercially usable licences, with a reported gain of more than a full day of cyclone forecasting lead
ByteDance launches SeedRealtime
SeedRealtime is a native audio-visual full-duplex model that decides for itself when to speak, reported to halve conversational pacing problems against cascaded systems, with benchmark detail and developer access still
Meta releases Muse Spark 1.2 and Muse Code
Muse Spark 1.2 is a coding-focused checkpoint with a 1M token context at unchanged standard pricing, shipped together with the terminal agent it was co-trained with.
NVIDIA releases Alpamayo 2 Super for autonomous driving
NVIDIA released Alpamayo 2 Super, a 34-billion-parameter open driving foundation model, licensed for commercial use, that outputs a trajectory, a causal explanation and a meta-action in one pass.
Alibaba releases Qwen3.8-Max
Qwen3.8-Max is Alibaba's flagship Qwen release for 2026, made generally available on August 3, 2026: a 2.4 trillion parameter Mixture-of-Experts model with 95 billion active parameters, a 1 million token
DeepSeek releases DeepSeek V4-Flash-0731
DeepSeek-V4-Flash-0731 is DeepSeek's official V4-Flash release, published on July 31, 2026: a 284 billion parameter MoE with 13 billion active parameters, a 1 million token context window, MIT-licensed weights and
AI knowledge base: what it is and what it delivers
Enterprise teams lose hours every week searching across fragmented apps and document drives. An AI knowledge base solves this by unifying corporate data into a central intelligence layer, delivering instant,
MiniMax releases MiniMax H3
MiniMax H3 is MiniMax's omni-modal generation model, released on July 31, 2026, generating up to 15 seconds of 2K video with native stereo audio from a unified text, image, video
Black Forest Labs releases FLUX 3
FLUX 3 is Black Forest Labs' new multimodal frontier model, announced on July 23, 2026, that generates 20-second video with synchronized audio and extends to image generation and robotic action
MiniMax: What is it and why is it important?
MiniMax is developed by Moonshot AI. It has quickly caught up to frontier model performance at only a fraction of the cost, and completely open weights. In this article we
Google releases Gemini 3.6 Flash
Gemini 3.6 Flash is Google's new default Gemini model, released on July 21, 2026 with a 1 million-token context window, a March 2026 knowledge cutoff and lower per-token pricing than
ByteDance releases Seed 2.1 Pro and Seed 2.1 Turbo
ByteDance has put a price aggressive, agent focused alternative to Claude Opus 4.6 in front of enterprise buyers, with independent benchmark scrutiny still to come.
OpenAI unveils Jalapeño: AI inference chip
Jalapeño marks OpenAI's expansion from models and products into custom silicon, giving the company a purpose-built inference chip co-developed with Broadcom and targeted for deployment by the end of 2026.
Sakana AI launches Fugu and Fugu Ultra
Sakana Fugu and Fugu Ultra are a multi-model orchestration API released by Sakana AI with a 1 million token context window and benchmark scores that the company reports as matching
Zhipu AI releases GLM-5.2
GLM-5.2 is Zhipu AI's open-weight Mixture-of-Experts model with a 1 million token context window, an MIT license, and coding benchmark scores that rival GPT-5.5 at a fraction of the cost.
SpaceX acquires Cursor maker Anysphere
SpaceX's 60 billion dollar all-stock acquisition of Anysphere, the maker of Cursor, is xAI's largest move into developer software to date and its first major entry into AI coding tools.
AI Automation: Where to start and what it delivers
AI automation works best when you start with one painful, repetitive process, automate it end to end, and measure the result. This practical 2026 guide shows where to begin, which
Apple rebuilds Siri on a custom 1.2-trillion-parameter Gemini model
Apple's rebuilt Siri AI, powered by a custom 1.2-trillion-parameter Google Gemini model inside Private Cloud Compute, is the company's clearest attempt yet to bring Siri level with frontier assistants.
Google releases DiffusionGemma 26B-A4B
DiffusionGemma 26B-A4B is Google DeepMind's experimental open text diffusion model, generating 256-token blocks in parallel for up to 4x faster output at the cost of a few benchmark points against
